gis projects for civil engineeing students

7/31/2019 Gis Projects for Civil Engineeing Students

1/25

gis projects for civil engineeing students

Environmental Modelling & Software

Volume 22, Issue 7, July 2007, Pages 951965

IRA-WDS: A GIS-based risk analysistool for

water distribution systems

K. Vairavamoorthya,b, , [Author Vitae], Jimin Yana[Author Vitae], Harshal M. Galgalea[Author Vitae], Sunil D. Gorantiwara[Author Vitae] a Department of Civil and Building Engineering, Loughborough University, LE11 3TU,

UK

b Department of Urban Water and Sanitation, UNESCO-IHE Institute for WaterEducation, P.O. Box 3015, 2601 DA, Delft, The Netherlands

Received 16 June 2005. Revised 30 March 2006. Accepted 30 May 2006. Availableonline 13 December 2006.

http://dx.doi.org/10.1016/j.envsoft.2006.05.027 ,How to Cite or Link Using DOI Cited by in Scopus (6) Permissions & Reprints

Abstract
http://www.sciencedirect.com/science/journal/13648152http://www.sciencedirect.com/science/journal/13648152http://www.sciencedirect.com/science/journal/13648152/22/7http://www.sciencedirect.com/science/journal/13648152/22/7http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff2http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff2http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff2http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt1http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt1http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt1http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt2http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt2http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt2http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt3http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt3http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt3http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt4http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt4http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt4http://dx.doi.org/10.1016/j.envsoft.2006.05.027http://dx.doi.org/10.1016/j.envsoft.2006.05.027http://www.sciencedirect.com/science/help/doi.htmhttp://www.sciencedirect.com/science/help/doi.htmhttp://www.sciencedirect.com/science/help/doi.htmhttp://www.sciencedirect.com/science?_ob=RedirectURL&_method=outwardLink&_partnerName=656&_eid=1-s2.0-S1364815206001538&_pii=S1364815206001538&_origin=article&_zone=art_page&_targetURL=http%3A%2F%2Fwww.scopus.com%2Finward%2Fcitedby.url%3Feid%3D2-s2.0-33846862440%26partnerID%3D10%26rel%3DR3.0.0%26md5%3D2a759fc2b96c2232aeca5c545a57f7f9&_acct=C000238598&_version=1&_userid=11019620&md5=9ed86c7ed500cbb1666c9ebe308c934bhttp://www.sciencedirect.com/science?_ob=RedirectURL&_method=outwardLink&_partnerName=656&_eid=1-s2.0-S1364815206001538&_pii=S1364815206001538&_origin=article&_zone=art_page&_targetURL=http%3A%2F%2Fwww.scopus.com%2Finward%2Fcitedby.url%3Feid%3D2-s2.0-33846862440%26partnerID%3D10%26rel%3DR3.0.0%26md5%3D2a759fc2b96c2232aeca5c545a57f7f9&_acct=C000238598&_version=1&_userid=11019620&md5=9ed86c7ed500cbb1666c9ebe308c934bhttp://www.sciencedirect.com/science?_ob=RedirectURL&_method=outwardLink&_partnerName=936&_eid=1-s2.0-S1364815206001538&_pii=S1364815206001538&_origin=article&_zone=art_page&_targetURL=https%3A%2F%2Fs100.copyright.com%2FAppDispatchServlet%3FpublisherName%3DELS%26contentID%3DS1364815206001538%26orderBeanReset%3Dtrue&_acct=C000238598&_version=1&_userid=11019620&md5=8173b5f227283b73b5e35b79f776c94fhttp://www.sciencedirect.com/science?_ob=RedirectURL&_method=outwardLink&_partnerName=936&_eid=1-s2.0-S1364815206001538&_pii=S1364815206001538&_origin=article&_zone=art_page&_targetURL=https%3A%2F%2Fs100.copyright.com%2FAppDispatchServlet%3FpublisherName%3DELS%26contentID%3DS1364815206001538%26orderBeanReset%3Dtrue&_acct=C000238598&_version=1&_userid=11019620&md5=8173b5f227283b73b5e35b79f776c94fmailto:[email protected]://www.sciencedirect.com/science/article/pii/S1364815206001538mailto:[email protected]://www.sciencedirect.com/science/article/pii/S1364815206001538mailto:[email protected]://www.sciencedirect.com/science/article/pii/S1364815206001538mailto:[email protected]://www.sciencedirect.com/science/article/pii/S1364815206001538http://www.sciencedirect.com/science?_ob=RedirectURL&_method=outwardLink&_partnerName=936&_eid=1-s2.0-S1364815206001538&_pii=S1364815206001538&_origin=article&_zone=art_page&_targetURL=https%3A%2F%2Fs100.copyright.com%2FAppDispatchServlet%3FpublisherName%3DELS%26contentID%3DS1364815206001538%26orderBeanReset%3Dtrue&_acct=C000238598&_version=1&_userid=11019620&md5=8173b5f227283b73b5e35b79f776c94fhttp://www.sciencedirect.com/science?_ob=RedirectURL&_method=outwardLink&_partnerName=656&_eid=1-s2.0-S1364815206001538&_pii=S1364815206001538&_origin=article&_zone=art_page&_targetURL=http%3A%2F%2Fwww.scopus.com%2Finward%2Fcitedby.url%3Feid%3D2-s2.0-33846862440%26partnerID%3D10%26rel%3DR3.0.0%26md5%3D2a759fc2b96c2232aeca5c545a57f7f9&_acct=C000238598&_version=1&_userid=11019620&md5=9ed86c7ed500cbb1666c9ebe308c934bhttp://www.sciencedirect.com/science/help/doi.htmhttp://dx.doi.org/10.1016/j.envsoft.2006.05.027http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt4http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt3http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt2http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/article/pii/S1364815206001538#vt1http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff2http://www.sciencedirect.com/science/article/pii/S1364815206001538#aff1http://www.sciencedirect.com/science/journal/13648152/22/7http://www.sciencedirect.com/science/journal/13648152


2/25

This paper presents the development of a new software tool IRA-WDS. This GIS-based software

predicts the risks associated with contaminated water entering water distribution systems fromsurrounding foul water bodies such as sewers, drains and ditches. Intermittent water distribution

systems are common in developing countries and these systems are prone to contamination when

empty. During the non-supply hours contaminants from pollutionsources such as sewers, open

drains and surface water bodies enter into the water distribution pipes through leaks and cracks.Currently there are no tools available to help engineers identify the risks associated with

contaminant intrusion into intermittent water distribution systems. Hence it is anticipated that

IRA-WDS will find wide application in developing countries. The paper summarises the detailsof the mathematical models that form the basis of IRA-WDS. It also describes the software

architecture, the main modules, and the integration with GIS using a tight coupling approach. A

powerful GUI has been developed that enables data for the models to be retrieved from thespatial databases and the outputs to be converted into tables and thematic maps. This is achieved

seamlessly through DLL calling functions within the GIS. This paper demonstrates the

application of the software to a real case study in India. The outputs from IRA-WDS are risk

maps showing the risk of contaminant intrusion into various parts of the water distribution

system. The outputs also give an understanding of the main factors that contribute to the risk.

Keywords

Water supply; Risk assessment; Contaminant intrusion; Developing countries; Intermittent water supply; GIS; Tight coupling

1. Introduction

Contamination of drinking water due to exposure to biological and chemical pollutants is a major

cause of illness and mortality. Recent evidence has demonstrated that external contaminant

intrusion into a water distribution network may be more frequent and of a greater importancethan previously suspected (LeChevallier, 1999). The problem of external contaminant intrusion

is more aggravated in developing countries where the pollutionsources crisscross with water

distribution systems and intermittent water supplies are prevalent ( [Choe and Varley, 1997],

[Seckler et al., 1998]and[Rosegrant et al., 2002]).

A serious problem arising from intermittent supplies, which is generally ignored, is the

associated high level of contamination which occurs in networks where there are prolongedperiods of interruption of supply due to negligible or zero pressure in the system. Such problems

lead to increased health risks as water becomes contaminated with pathogens due to the intrusion

from surrounding foul water bodies (e.g. sewers, ditches), through joints and cracks in the
http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib33http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib33http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib33http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib13http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib13http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib13http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib41http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib41http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib39http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib39http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib39http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib39http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib41http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib13http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib33


3/25

deteriorated water distribution pipes. Hence in developing countries the distribution network has

become a point at which contamination frequently occurs to unacceptably high levels, posing athreat to public health ( [Swerdlow et al., 1992]and[Besser et al., 1995]).

Currently there are no software tools available to help engineers identify the risks associated with

contaminant intrusion into intermittent water distribution systems. The need for such tools hasresulted in the development of the Improved Risk Assessment of Water Distribution Systems

(IRA-WDS) tool. The application of this tool will enable engineers to better manage waterquality by developing appropriate control measures to minimise these risks by, for example,

prioritising their operational maintenance strategies.

Previous studies mainly focussed on risk due to breakage of pipes. All these risk assessment

methodologies are based on the principle of regression ( [Clark et al., 1982],[Shamir and

Howard, 1979],[Walski, 1987]and[Walski and Pelliccia, 1982]) or probability analysis (

[Andreou et al., 1987a],[Andreou et al., 1987b],[Deb et al., 1998],[Eisenbeis et al., 1999],[Herz, 1996],[Herz, 1998]and[Lei, 1997]) and need large amounts of historical data. But the

data needed for this purpose is extremely rare and insufficient ( [Ang and Tang,1984]and[Guymon and Yen, 1990]). Moreover these studies were developed for the waterdistribution systems which operate at high pressures where the problem is one of leakage rather

than contaminant intrusion. But, as stated earlier, pressures in water distribution systems in

developing countries are often negligible due to intermittent water supply and contaminants mayenter the system through the deteriorated pipes during the periods of low pressure. As the

techniques developed for risk assessment due to breakage do not highlight the contaminant

intrusion, these are not applicable for intermittent water distribution systems. The techniques

developed to consider contaminant intrusion under low or negative pressure resulting fromtransient use rely on the leakage or breakage data as a surrogate for the intrusion pathway (

[Boyd et al., 2004a],[Boyd et al., 2004b],[Funk et al., 1999]and[LeChevallier et al., 2003]).

But these methods have generally assumed that contaminant sources exist around all waterdistribution pipes and were applied to continuous systems.

Lindley and Buchberger (2002)developed a technique to identify the locations in the waterdistribution systems that may be susceptible to unintended contaminant intrusion. They argued

that three susceptibility conditions must be met for an intrusion into a water distribution system

to occur. These are: an adverse pressure gradient, an intrusion pathway and a contaminantsource. However, the technique developed by Lindley and Buchberger simply performs a spatial

analysis to produce a combination of the three susceptibility conditions and does not model the

susceptibility conditions for contaminant intrusion.

Unlike the previous methods, the IRA-WDS assesses the risk of contaminant intrusion into the

water distribution system by modelling the process of contaminant transport from

pollutionsources such as sewers, open drains and foul water bodies. In addition it includes apipe-condition assessment model that is used to indicate intrusion pathways. This model uses

over 20 indicators to estimate pipe condition.

Risk assessment of water distribution systems due to pollutionsources is a spatial process as the

risk components such as hazard(due to pollutionsources) and vulnerability (due to the water
http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib44http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib44http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib44http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib9http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib9http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib9http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib14http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib14http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib14http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib42http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib42http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib42http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib42http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib46http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib46http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib46http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib47http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib47http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib47http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib1http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib1http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib1http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib1http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib2http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib2http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib2http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib15http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib15http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib15http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib16http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib16http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib16http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib29http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib29http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib30http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib30http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib30http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib35http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib35http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib35http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib3http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib3http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib3http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib3http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib27http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib27http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib27http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib10http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib10http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib10http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib10http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib11http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib11http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib11http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib22http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib22http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib22http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib34http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib34http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib34http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib36http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib36http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib36http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib34http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib22http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib11http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib10http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib10http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib27http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib3http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib3http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib35http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib30http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib29http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib16http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib15http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib2http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib1http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib1http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib47http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib46http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib42http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib42http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib14http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib9http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib44


4/25

distribution system) are both spatially diverse and need large volumes of spatially diverse data.

The pollutionsources such as the sewer system, open drains and surface water bodies are parts ofthe urban sanitation infrastructure and geo-spatially referenced. In addition, the vulnerability

(condition in terms of deterioration) of the water distribution system is spatially distributed since

the deterioration attributes are spatial in nature. The risk of contaminant intrusion into a water

distribution system is a product of the interaction between the hazard and vulnerabilitycomponents and is therefore a spatial process. Hence it is necessary to know the spatial outputs

in the form of maps representing the degree of severity of the associated risk to the water

distribution system so that decision makers can draw up priorities for the rehabilitation of thewater distribution system.

Geographical information systems (GIS) are powerful tools for handling spatial data, performing

spatial analysis and manipulating spatial outputs. A GIS also provides a consistent visualisation

environment for displaying the input data and results of a model. This ability of GIS is very

useful in a decision-making process. The integration of GIS and external models enables theutilisation of the advantages of both ( [Goodchild et al., 1992],[Goodchild et al., 1993],

[Goodchild et al., 1996],[Fotheringham and Rogerson, 1994],[Fischer et al., 1996],[Longleyand Batty, 1996],[Fotheringham and Wegener, 2000]and[Argent, 2004]). Since GIS allows theinput of spatial data into a model and provides the outputs in spatial forms, it was decided tointegrate a GIS into the tool developed for risk assessment. The combination of GIS and

modelling tools enables the conversion of large amounts of data into information and then into

practical knowledge that is useful for risk assessment. The intrinsic ability of GIS to store,analyze and display large amounts of spatial data enables it to make a significant contribution to

risk assessment.

This paper presents the development of a new software tool: the Improved Risk Assessment of

Water Distribution Systems (IRA-WDS). The software consists of three models: the contaminant

ingress model, the pipe-condition assessment model and the risk assessment model. The modelsare developed in the C++ language and integrated with the GIS using a tight-coupling approach.

This paper highlights the different integration methodologies available for the integration of

environmental models with GIS and describes the integration methodology used for developing

the IRA-WDS; the modelling tools of IRA-WDS and their integration with GIS. The uses ofIRA-WDS and a case study of its application are described.

2. Method

This section presents the methodology used for the improved risk assessment for water

distribution systems due to contaminant intrusion. Three risk factors have been identified for

contaminant intrusion into a water distribution system. These are: the section of pipe in thecontaminant zone (SPCZ), the contaminant concentration and the condition of the water

distribution pipes in terms of deterioration. These risk factors are established by the associated

models, namely, contaminant ingress (contaminant zone and contaminant seepage) and pipecondition assessment models. The three risk factors obtained from these models are combined to

obtain a risk index for each pipe by the use of a risk assessment model.
http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib25http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib25http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib25http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib26http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib26http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib20http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib20http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib20http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib19http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib19http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib19http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib37http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib37http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib37http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib37http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib21http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib21http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib21http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib4http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib4http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib4http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib4http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib21http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib37http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib37http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib19http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib20http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib26http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib25http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24


5/25

Fig. 1presents the conceptual model for IRA-WDS. Contaminants from the pollutionsources

(e.g. leaky sewers, drains) seep through the soil and percolate down. In this process, thesepollutionsources develop a contamination zone (CZ) and if all or part of a water distribution

system's pipe work passes through the contamination zone, contaminants may find entry into the

pipes through such entry points as cracks and leaky joints. Hence, there is a risk of contaminant

intrusion if a water pipe lies in a contaminant zone and at the same time the condition of the pipehas been compromised so as to allow contaminant entry.

Fig. 1. Conceptual model for IRA-WDS.

Thus the methodology consists of developing and integrating the following three separate modelsas shown inFig. 2.
http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig1http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig1http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig2http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig2http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig2http://www.sciencedirect.com/science/article/pii/S1364815206001538http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig2http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig1


6/25

Fig. 2. Components of IRA-WDS modelling tools.

2.1. Contaminant ingress

The contaminant ingress model estimates the contaminant zone of the pollutionsources (sewer

pipes, open drains/canals and surface water bodies), and identifies the section of the water

distribution pipes passing through the contaminant zone (SPCZ). This model consists of two

parts: the contaminant zone and contaminant transport.

The contaminant zone model predicts the envelope of pollution emanating from pollutionsources:the contaminant zone. The model is based on the seepage process, which forms part of the theory

of soil mechanics. It is assumed that the seepage of contaminants from pollutionsources such as

unlined canals/drains and surface water bodies follow a saturated flow while that from

pollutionsources such as sewer pipelines and lined canals/drains follows an unsaturated flow.The separate sets of equations for the estimation of a contaminant zone were developed for

saturated and unsaturated flows by modifying the approach presented byHarr (1962). The details

of the development of these equations can be found inYan et al. (2002). Once the contaminant

zones formed due to pollutionsources are established, the model uses spatial techniques toestimate the sections of the water distribution pipes that intersect with the contaminant zone

(SPCZ). This is achieved using geometric algorithms. The algorithms calculate the length ofcontaminated pipe (LC) in the contaminant zone using the upstream and downstream points ofintersection between the contaminant zone and the segment of the water pipe passing through the

contaminant zone, as given in Eq.(1):

(1)
http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib28http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib28http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib28http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib48http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib48http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib48http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd1http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd1http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd1http://www.sciencedirect.com/science/article/pii/S1364815206001538http://www.sciencedirect.com/science/article/pii/S1364815206001538http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd1http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib48http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib28


7/25

where LCk is the length of the kth pipe in the contaminant zone (m); upkand dpkare the upstream

and downstream intersection points of the kth pipe with the contaminant zone and NP is thenumber of water distribution pipes.

The contaminant transport model simulates the water flux and the variable concentration of the

contaminants within the contaminant zone and then predicts the contaminant loading on theSPCZ. This is done by modelling the contaminant transportation process through the soil from

the pollutionsources towards the water distribution pipes. Two separate models were developedfor estimation of the water flux: the unsaturated flow model for pollutionsources such as sewer

pipes and lined drains/canals and the saturated flow model for unlined drains/canals and surface

water bodies.

For the saturated flow model the flow region is divided into a flownet and for each cell of the

flownet the pore-water velocity vi is estimated by combining the velocity potential and Darcy's

equation (Fetter, 1999). For the unsaturated flow model, the pore water velocity is calculatedusing the water flux obtained from the GreenAmpt model and the projected water content

(Green and Ampt, 1911).

These pore-water velocities are then used by the contaminant transport model to calculate the

variable contaminant concentration in soil. In this paper, an analytical solution for the advection-

diffusion-reaction equation developed byBear (1972)is used, as given in Eq.(2):

(2)

where

where RC is the relative concentration, tis time (hours),z is the depth along the flow path (cm),

D is the dispersion coefficient (cm2/day), is the pore-water velocity (cm/h), b is the bulk

density (g/cm3); Kd is a sorption constant, n is the porosity, S is the solid-phase concentration

(mg/l), and is the first-order decay coefficient in the liquid phase (1/h).

The contaminant transport model estimates the profile of the contaminant concentration in thesoil. The contaminant concentration along the SPCZ is the average concentration at its start and

end intersection points as given by Eq.(3).

(3)
http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib18http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib18http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib18http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib23http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib23http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib23http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib8http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib8http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib8http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd2http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd2http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd2http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd3http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd3http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd3http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd3http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd2http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib8http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib23http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib18


8/25

where CCkis the average contaminant concentration along SPCZ of pipe k,RCkup is the

contaminant concentration at the upstream intersection point andRCkud is the contaminantconcentration at the downstream intersection point of pipe k.

2.2. Pipe condition assessment

The pipe condition assessment (PCA) model estimates the relative condition of the pipes in the

water distribution network to assess the potential intrusion pathway. The condition of each pipeis assessed by means of numerous factors related to the physical, environmental and operational

aspects of a water distribution system (seeTable 1). These factors are grouped into different

indicators at three levels, depending on the nature of the influence of each factor on the

deterioration process of the pipe. The uncertainties inherent in these pipe condition indicators aredescribed using fuzzy set theory (Zadeh, 1965).

Table 1. Pipe indicators and their groups at different levels

Description Level 1 Level 2 Level 3 FinalHazenWilliam coefficient of

friction (C) is considered to

characterise this influence

Material decay

Pipe indicators

Physical

indicatorsPCA

Larger diameter pipes are lessprone to failure than smaller

diameter pipes

Diameter (mm)

Larger length pipes are more

prone to failure than smallerlength pipes

Length (m)

The pipes having internalprotection by lining and/or

coating are less susceptible tocorrosion

Internal

protection

The pipes having external

protection by lining and/or

coating are less susceptible todeterioration

External

protection

Improper bedding may result

in premature pipe failure

Bedding

condition

Installationindicators

Poor workmanship may

deteriorate the pipes andcause more risk regardless of

pipe age and other factors

Workmanship

Some types of jointsexperience premature failure

(e.g. leadite joints)

Joint method

The more the joints a pipe No. of joints
http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl1http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl1http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl1http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib50http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib50http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib50http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib50http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl1


9/25

Description Level 1 Level 2 Level 3 Final

has, the more risk of the pipegetting structurally worse

The effects of pipe

degradation become more

apparent over time

Year of

installation

Corrosion

indicators

Environmental

indicators

Pipe deteriorates quickly in

more corrosive soil and the

degree of deteriorationdepends on the pipe material

Soil corrosivity

(Ohm)

The more permeable surface

allows more moisture to

percolate to the pipe. Thesurface salts will be carried to

the pipe with moisture

Surface

permeability

The water pipes aredeteriorated by thegroundwater table

Gouundwater

condition

Pipes buried at higher depth

have more possibility of

failure than those buried atshallower depths

Buried depth

(m)Load/strength

indicators;intermittency

indicators

Pipe failure rate increases

with traffic loadsTraffic load

Changes to internal waterpressure will change stressesacting on the pipe

Maximum

pressure

The more valves, the more is

the deterioration of the pipeNo. of valves

Operational

indicators

The more water supplies, themore the pipes will be

deteriorated

No. of watersupply/day

The more the duration ofwater supply, the less chances

of pipe failure

Duration ofwater

supply/day

The number of pipe breakages

per year

Breakage

history Failure indicator

Based on their similarities, the first-level indicators are aggregated to form the second-levelindicators. Similarly, the second-level pipe condition indicators are aggregated to form the final

indicator. Based on the hierarchical pipe condition structure established from the above

aggregation process, fuzzy composite programming ( [Bardossy and Duckstein,

1992]and[Bender and Simonovic, 2000]) is used to compute an indicator distance metric for

each indicator, and finally an overall distance metric is obtained using Eq.(4).
http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib5http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib5http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib5http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib5http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib7http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib7http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib7http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd4http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd4http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd4http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd4http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib7http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib5http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib5


10/25

(4)

whereLj is the distance metric of pipej, wj,i is the weighting of indicator i in groupj,pj is a

balance factor among the indicators for groupj,fbj,j is the best value for indicator i in groupj,fwj,i

is the worst value for indicator i in groupj, andfj,i is the actual value for indicator i in groupj.

The weightings are assigned to indicate the relative importance of the various pipe indicators in aparticular group. They are generated by following an analytical hierarchy process (AHP) (Saaty,

1980). Balance factors that determine the degree of compromise between indicators of the same

group are assigned to different groups.

The final overall distance metric is a fuzzy number represented by a membership function and is

defuzzified using the maximising and minimising sets (Chen, 1985). The defuzzified numbersthus rank the pipes according to their conditions. This metric will be used as a surrogate for the

vulnerability of the water distribution pipe (Eq.5). The details can be found inYan and

Vairavamoorthy (2003).

(5)

where VUk is the vulnerability of the water distribution pipe k, df is the method for

defuzzification and TFk is the trapezoidal fuzzy number for pipe k.

2.3. Risk assessment

The risk of contaminant intrusion into the water distribution system results from the interaction

between the hazard agent and the vulnerability of the water distribution pipe. The riskassessment model calculates this risk of contaminant intrusion by combining the outputs from

the above two models.

In this study, the hazard agent is the contaminant loading along the section of pipes in the

contaminant zone. The contaminant loading is obtained from the length of the section of pipe in

the contaminant zone and the concentration of contaminants along this section of pipe (Eq. (6)).As stated earlier, these parameters (SPCZ and the contaminant concentration) are obtained with

the help of the contaminant ingress model.

(6)

where HAk is the estimation of hazard for pipe k, LCkis the normalised length of the polluted

pipe k, obtained from Eq.(1),rk is the normalised radius of pipe kand CCkis the normalisedcontaminant concentration of pipe k, obtained from Eq.(3).

The condition of the water distribution pipe, which is calculated with the help of the pipecondition assessment model (Eq.(5)), is used as a surrogate of the water pipe's vulnerability to

intrusion.
http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib40http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib40http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib40http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib40http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib12http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib12http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib12http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd5http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib49http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib49http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib49http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib49http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd6http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd6http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd6http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd1http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd1http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd1http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd3http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd3http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd3http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd3http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd1http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd6http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib49http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib49http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd5http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib12http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib40http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib40


11/25

The combination of hazard agent and the vulnerability of the water pipe produces the risk index

for contaminant intrusion into the water distribution system. This paper uses a conventionalmultiple criteria evaluation (MCE), the weighted linear combination (WLC), to calculate the risk

index.

(7)

where RI is the risk index, wh is the weighting for hazard agent, wv is the weighting for thevulnerability of the water pipe, VUk is the vulnerability of the water pipe, obtained from Eq.(5)

and HAk is obtained from Eq.(6).

3. Software development

The three models were developed in the C++ programming language and integrated with

ArcView 3.2 GIS, using a tight-coupling approach (achieved with the Avenue programminglanguage and dynamic link libraries (DLLs)). This GIS-based tool is the spatial decision support

system, which has been named as the Improved Risk Assessment-Water Distribution System(IRA-WDS).

ArcView Dialog Designer has been used to design the graphical user interfaces that enable

seamless interaction of the user with the models developed. The interface allows the input andretrieval of data to and/or from the model in a user-friendly way. The GUI also provides the

functionalities for generating input files, running external models and converting and displaying

model outputs from formats such as ASCII files to thematic maps.

The section below describes the main components of the software. The software includes several

modules based on mathematical models and a GIS-based interface.

3.1. Contaminant ingress model

The contaminant ingress model consists of three modules: the contaminant zone, contaminantseepage, and contaminant transport. The data required for this model includes water distribution

system data, pollutionsource data (sewer, canal, and foul water bodies) and soil data. The details

of modules are described inTable 2.

Table 2. Module descriptions for contaminant ingress model

Module Purpose Input Output

Contaminant

zone

To determine the contaminant

zone of pollutionsources andSPCZ along water distribution

pipe. SPCZ is one of the risk

factors in risk assessment model

Networks data for

pollutionsources and

water distributionsystem

SPCZ along

water

distribution pipe(m)

Contaminantseepage

To calculate flow velocity due tothe pollutionsources. The flow

Properties for soil andpollutionsources

Pore-watervelocity (cm/h)
http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd6http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd6http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd6http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl2http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl2http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl2http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl2http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd6http://www.sciencedirect.com/science/article/pii/S1364815206001538#fd5


12/25


velocity is used for contaminanttransport model

Contaminanttransport

To calculate contaminant

concentration above water

distribution system. Theconcentration is another factors in

risk assessment model

Properties for soil andpollutionsources

Relative

contaminantconcentration

The contaminant zone module predicts the contaminant zone for a particular pollutionsource

(sewers, open drains and foul water bodies). The SPCZ for the water distribution system is thenidentified using spatial analysis involving contaminant zones and water distribution pipes.

The contaminant seepage module predicts flow velocity for both saturated and unsaturatedflows, using Darcy's law and the Green Ampt model respectively. The flow velocities are inputs

for the contaminant transport module.

The contaminant transport module calculates the contaminant concentration along the SPCZ for

steady and unsteady state conditions. These outputs represent the hazard factors in the risk

assessment model.

3.2. Pipe condition assessment model

The pipe condition assessment model consists of five modules: the fuzzy calculator; criteria

normalisation; the weighting generator; the fuzzy composite programme; and the classification

module. The data required for the pipe condition assessment model concerns the water

distribution system, the pipe deterioration indicator along with groundwater and pressure zonedata. The details of these modules are described inTable 3.

Table 3. Module descriptions for pipe condition assessment model


Fuzzy

calculatorTo perform fuzzy arithmetic

Fuzzy membership

functions

Aggregated fuzzy

number

Criteria

normalisation

To normalise deterioration

factors to the same scale and

used for composite

programming

Properties for soil

and pollutionsources

Normalised

criteria for

composite

programming

Weight

generator

To generate weights for

deterioration factors usingeither analytical hieratical

process (AHP) or assign

weight directly and balance

factors

Pair-wise

comparison from

interviewing experts

Weight and

balance factor
http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl3http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl3http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl3http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl3


13/25


Fuzzycomposite

programming

To calculate condition index

using deterioration indicators

Pipe deteriorationdata for water

distribution system

Pipe condition

index

Ranking andclassification

Rank pipe condition index that

are fuzzy numbers and classifyinto groups given by user

Pipe condition index

from compositeprogramming

Condition groupclassifications

Thefuzzy calculator module performs fuzzy arithmetic calculations. Both triangular and

trapezoidal fuzzy numbers are considered by this module. Various operators (e.g. addition,

subtraction, multiplication, division, power), are overloaded so that normal arithmetic operators

(+, , , /) can be used.

The criteria normalisation module normalises pipe deterioration factors using criteria A, B, C orD that are used in the composite programming.

The weighting generator module generates weightings and balance factors. The weights andbalance factors (used for pipe condition indictors and groups) are either assigned directly by the

user or generated using the analytical hieratical process (AHP). In addition, default weightings

obtained by interviewing engineers and practitioners working on network operations have beengiven.

Thefuzzy composite programming module aggregates the normalised deterioration indicators,weightings and balance factors to calculate a pipe condition index for each pipe. Each class in

this module performs this operation for a specific level of the composite hierarchy structure.

The ranking and classification module defuzzifies the fuzzy pipe-condition index, obtained from

the fuzzy composite programming module for each pipe and ranks/groups pipes according to

their defuzzified pipe-condition index.

3.3. Risk assessment model

The risk assessment model consists of four modules: risk factors normalisation; weightinggenerator; risk estimator and the classification module. The details of these modules are

described inTable 4.

Table 4. Module descriptions for risk assessment model


Criteria

normalisation

To normalise risk factors to the

same scale and used for risk

estimator

Outputs from ingress

model and pipe

condition assessmentmodel

Normalised risk

factors

Weight

generator

To generate weights for risk

factors using either analytical

Pair-wise comparison

from interviewing

Weight for risk

factors
http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl4http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl4http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl4http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl4


14/25


hieratical process (AHP) orassign weight directly

experts

Risk estimator

To generate risk index using

the risk factors obtained from

ingress model and pipecondition assessment model

Normalised risk factors Risk index

Ranking and

classification

Rank pipe risk index and

classify into groups given byuser

Risk index from risk

estimator

Risk group

classifications

The risk factor normalisation module normalises a particular risk factor. For example the SPCZ

and contaminant concentration risk factors are normalised and used to calculate the hazard. The

hazard and vulnerability risk factors are then normalised and used to assess risk.

The weighting generator module generates weightings for each of the risk factors. These areeither assigned directly by the user or generated using the analytical hieratical process (AHP).

The risk estimator module generates the risk index for each pipe by using a weighted linear

combination (WLC).

The ranking and classification module ranks and groups pipes according to their risk index (the

number of groups are specified by the user).

3.4. Integration into the GIS

The integration of environmental models with GIS has been discussed by several researchers ([Goodchild et al., 1992],[Goodchild et al., 1993],[Stuart and Stocks, 1993],[Batty and Xie,

1994],[Fotheringham and Rogerson, 1994],[Fischer et al., 1996],[Goodchild et al., 1996],

[Karimi and Houston, 1996],[Longley and Batty, 1996]and[Fotheringham and Wegener,2000]). In addition,Karimi and Houston (1996)andTait et al. (2004)also attempted to classify

different integration methodologies. In general, three methods of integration or coupling are used

to link environmental models with GIS. These are: loose coupling, tight coupling and embeddedcoupling. These three integration methods differ in their architectural characteristics and this

depends on the degree and form of data exchange or sharing between the GIS and the external

models ( [Goodchild et al., 1992],[Goodchild et al., 1993],[Nyerges, 1992]and[Fedra, 1993]).

With loose coupling, the GIS serves as both a pre-processor and a postprocessor to the modelling

system, while tight coupling is the integration of the models and GIS under a common interface.In embedded coupling, the models are developed within the GIS environment or alternatively a

GIS component is added to the modelling system (Huang and Jiang, 2002). Where complex

models of different domains are to be integrated with GIS, tight coupling is needed (Karimi andHouston, 1996). The tight coupling approach also offers full control to the experienced user and

minimal interaction to the novice. According toTait et al. (2004), through tight coupling,

modellers can spend more of their effort on building the models themselves. Therefore the tight
http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib25http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib25http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib25http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib43http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib43http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib43http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib6http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib6http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib6http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib6http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib20http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib20http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib20http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib19http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib19http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib19http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib26http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib26http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib26http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib37http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib37http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib37http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib21http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib21http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib21http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib21http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib45http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib45http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib45http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib25http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib25http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib25http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib38http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib38http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib38http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib17http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib17http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib17http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib31http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib31http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib31http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib45http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib45http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib45http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib45http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib31http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib17http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib38http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib25http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib45http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib21http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib21http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib37http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib32http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib26http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib19http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib20http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib6http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib6http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib43http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib25http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24http://www.sciencedirect.com/science/article/pii/S1364815206001538#bib24


15/25

coupling approach has been adopted in the development of IRA-WDS.Fig. 3illustrates the

integration of models with GIS.

Fig. 3. Integration of models with GIS.

The GUI enables the data to be retrieved from the spatial databases, and then passed to therequired models. The outputs from the models are then converted and presented as tables and

thematic maps. This is all achieved seamlessly through DLL calling functions within the GIS.

A Document Graphical User Interface (DocGUI) for IRA-WDS (Fig. 4) has been created based

on the View Document GUI of ArcView. Hence the GUI has similar functionalities to View

document, namely allowing the user to display, explore, query and analyze geographic data in

IRA-WDS.

Fig. 4. The major components of GIS-based IRA-WDS.

This IRA-WDS document GUI includes a selection of controls such as menus, buttons and tools

that can be used to interact within the IRA-WDS document GUI. The menus provide sub-menusfor generating and loading the input files, for executing the models (through DLL) and viewing

the model outputs in tabular or thematic map formats. The model outputs are stored in both

ASCII and shape file format (using Avenue script written in the IRA-WDS). The main menus inIRA-WDS launch various important dialog boxes:
http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig3http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig3http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig3http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig4http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig4http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig4http://www.sciencedirect.com/science/article/pii/S1364815206001538http://www.sciencedirect.com/science/article/pii/S1364815206001538http://www.sciencedirect.com/science/article/pii/S1364815206001538http://www.sciencedirect.com/science/article/pii/S1364815206001538http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig4http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig3


16/25

The data preparation menu launches dialog boxes (Fig. 5) that allow the user to convert

data from one format to another and to upload relevant shape files and tables into the

IRA-WDS platform.

Fig. 5. Addition of shape files through Data Preparationmenu.

The contaminant ingress menu launches dialog boxes (Fig. 6) that allow the uploading ofthe relevant shape-files for the model (i.e. water distribution network; sewer network;

canal network; surface water bodies and soil type (note that there is a default database of

soils)).
http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig6http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig6http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig6http://www.sciencedirect.com/science/article/pii/S1364815206001538http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig6http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig5


17/25

Fig. 6. The main interface for Contaminant Ingress Model.

Thepipe condition assessment menu launches several dialogs boxes (Fig. 7) that allowuploading of the relevant shape-files for the model (i.e. water distribution system, soil

data, groundwater table and hydraulic pressure). The user can specify the pipe indicatorsto be considered for the condition assessment and their respective crisp/fuzzy values(fuzzy values are inputted by means of a suitable membership function). Note that there

is a default database for pipe materials. The user can specify the weightings and balance

factors for the pipe indicators (note this dialog box will launch another that will allow the

weightings to be generated by the AHP method) and a local or global analysis and theninput the range of values to be used (minimum and maximum). Note that there is a

default database for the ranges.
http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig7http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig7http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig7http://www.sciencedirect.com/science/article/pii/S1364815206001538http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig7


18/25

Fig. 7. The main interface for Pipe Condition Assessment Model.

The risk assessment menu launches a dialog box (Fig. 8) that allows uploading of therelevant output files from the contaminant ingress and pipe condition assessment model.In addition it allows the user to specify the weights required for hazard and vulnerability

(note that this dialog box will launch another that will allow the weights to be generated

by the AHP method).


19/25

Fig. 8. The main interface for Risk Assessment Model.

4. Application of IRA-WDSIRA-WDS was applied to one of the ten zones of the Guntur Municipal Corporation, Guntur,

Andhra Pradesh, India. This zone is the B. R. Stadium Zone and called Zone VIII. This zone

has a population of about 60,000 and an area of 4 km2. As in most parts of the country, an

intermittent water supply system is prevalent in this area. Water is supplied through pipe

networks for about 1 h per day.

The drinking water supplies in this area are prone to contamination due to the various

pollutionsources that exist in this zone. Therefore the authorities responsible for managing this

system are keen to understand the risks to the water supply associated with the hazards described

above and to develop a maintenance strategy that will provide maximum improvements to waterquality within their limited budget.

At several locations, underground sewer pipes run close, parallel and above the water

distribution pipes. Frequent leaks are reported in the sewer network due to blockages. Sewer

pipes only cover a small part of the case study area. Most sewage is directly discharged into opendrains and canals. Considerable amounts of seepage occur from these open drains and, since they
http://www.sciencedirect.com/science/article/pii/S1364815206001538


20/25

are at ground level, this seepage can reach the water distribution pipes. Stagnant water is also

present in natural depressions in the study area.

4.1. Data collection

The application of IRA-WDS to the Guntur case study required the construction of severaldatabases. The details of the data requirements of IRA-WDS are given inTable 5. These

databases were constructed using data collected over a period of 12 months from June 2002 withthe assistance of the local authorities and a local NGO. Four thematic maps were constructed

related to: the water distribution network; the sewer network; the network of canals/open drains

and the surface water bodies (seeFig. 9).

Table 5. The data requirement of IRA-WDS

Water distribution

systems

Network map (link and node coordinates in metres) and for each

pipe of network: length of pipe, joint method, material type,

traffic load, surface type, internal protection, external protection,bedding condition, workmanship, diameter of pipe, installation

year, bury depth of start node, bury depth of start node, no of

connections, no of breaks per year, number of valves, duration ofwater supply per day (h/day), number of times water supplied per

day

Pollutionsources

The possible pollutionsources and the data needed for each

pollutionsources is described below. Underground sewer pipe:

network map (link and node coordinates), and for each pipe itslength, bury depth, material, leakage rate (m/day) and diameter.

Lined open ditch/drain: network map (link and node coordinates),

and for each ditch/drain of network, its length, material, depthand leakage rate. Unlined open ditch/drain: network map (link

and node coordinates), and for each ditch/drain of network, its

length, depth, soil type and seepage rate. Surface water bodies:surface water bodies map and for each surface water body, its

area, depth, soil type and seepage rate

Soil

Different soil types and for each soil type: saturated volumetricwater content, initial volumetric water content, saturatedhydraulic conductivity (cm/h), soil characteristic curve

coefficient, soil porosity, air entry head (cm), pore size index,

bulk density (g/cm3)and fraction organic content

Contaminant/pollutant

For each pollutionsource: liquid phase decay (/h), diffusion

coefficient (cm2/day), organic carbon partition coefficient of the

pollutant

Pipe material

Different types of pipe material used and for each type: corrosion

index, maximum pressure rating (kg/cm2), maximum load rating

(m-kg/m), design life, maximum diameter (mm), minimum

diameter (mm) and the variation of HazenWilliam coefficient
http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl5http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl5http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl5http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig9http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig9http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig9http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig9http://www.sciencedirect.com/science/article/pii/S1364815206001538#tbl5


21/25

(C) of friction with the age of pipe

Groundwater

Groundwater zones in the area and for each zone, average depth

to groundwater table (m) and average fluctuation of groundwater

(m)

Pressure zone

Pressure zones in the water distribution system network and for

each zone the pressure in the system

Fig. 9. Thematic maps for case study area in Guntur, India. (a) Water distribution system;

(b) sewer system; (c) canal/open drains system and (d) foul water bodies.

The three models of IRA-WDS (contaminant ingress, pipe condition assessment and risk

assessment model) extract the necessary information from the corresponding databases of eachtheme in order to run the simulations. The outputs of each model are returned back to the geo-

database where the corresponding output themes are generated.

In this study, three output themes are generated namely: the SPCZ theme generated from the

ingress model; the pipe condition theme generated from the pipe condition assessment model and

the risk assessment theme generated from the risk assessment model.

4.2. Results

The outputs of IRA-WDS (SPCZ, pipe condition and risk maps) for the Guntur case study areaare discussed in this section.

4.2.1. SPCZ Map

The SPCZ map shown inFig. 10, indicates the section of pipes in the water distribution system

that are within the contaminant zone of pollutionsources (sewers, canals/open drains and foulwater bodies). In addition, the relative contaminant loading on the SPCZ is also shown.

Therefore this output gives an indication of the sections of water pipes in danger of being

contaminated.


22/25


23/25

Fig. 11. Water pipe condition map for Guntur case study area.

4.2.3. Risk map for contaminant intrusion

The risk map generated for the Guntur case study area is shown inFig. 12. This figure indicates

that most of the water pipes have a medium or low risk. However, a small number of pipes havea high risk of contamination. From the risk map inFig. 12, several recommendations were made

to reduce the risk of contaminant intrusion. These suggested that the authorities should:

replace/rehabilitate AC pipes which are found to be in bad condition and hence have avery high susceptibility to contaminant intrusion (e.g. risk area A).

undertake a leakage detection and repair programme in areas with many joints andconnections (e.g. risk area B);

inspect open drains and reline where necessary (e.g. risk area C);
http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig12http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig12http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig12http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig12http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig12http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig12http://www.sciencedirect.com/science/article/pii/S1364815206001538http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig12http://www.sciencedirect.com/science/article/pii/S1364815206001538#fig12


24/25

provide protection to water pipes in areas where they are close to the open drains (risk

areas D);

de-water and fill foul water bodies in the north east (e.g. risk area E).

Fig. 12. Risk map for contaminant intrusion for Guntur case study area.

Though the results obtained for the case study area could not be validated with actual field data,

nevertheless the IRA-WDS provides engineers with valuable insights into the risk of watercontamination in a water distribution system. One of the major benefits of using IRA-WDS, is

that it is possible for the decision makers to gauge the impacts of the above recommendations on

the risk index. This can be achieved by simply modifying the database appropriately andrerunning the model. However, as there may be several other objectives related to an investmentstrategy for the water distribution system, these should also be considered. For example, it would

be prudent to combine the outputs of this model with a hydraulic model to establish the most

significant pipes in terms of both the risk of contaminant intrusion and changes required toimprove the hydraulic carrying capacity of the system.

5. Conclusions
http://www.sciencedirect.com/science/article/pii/S1364815206001538


25/25

This paper recognises the need of municipal engineers in developing countries for a tool to

assess the risk of contaminant intrusion into a water distribution system so as to prioritise riskmitigation actions. Currently there are no tools available to enable engineers to identify the risks

associated with contaminant intrusion into intermittent water distribution systems. This research

has identified a sourcepathwayreceptor approach for establishing the risk of contaminant

intrusion and developed a powerful tool, IRA-WDS for this purpose. This tool enables decision-makers to prioritise their investments in relation to water quality management.

IRA-WDS consists of three separate models. The contaminant ingress model establishes the

hazard by estimating the contaminant loading on a section of water distribution pipes passing

through a contaminant zone. The pipe condition assessment model establishes the vulnerabilityby estimating the relative condition of pipes using fuzzy composite programming techniques.

The risk assessment model then calculates risk by combining the outputs from the above two

models (hazard and vulnerability).

The three models were developed in the C++ programming language and integrated with GIS

using a tight-coupling approach (achieved with the Avenue programming language and dynamiclink libraries (DLLs)). IRA-WDS also has a powerful GUI that enables data for the models to beretrieved from the spatial databases and the outputs from the models to be converted and

presented as tables and thematic maps. This is all achieved seamlessly through DLL calling

functions within the GIS.

IRA-WDS has been applied to a water distribution network in south India and the outputs used to

generate several thematic maps. These maps identified sections of the system that were most atrisk and also provided an understanding of the main factors that contributed to the risk. The

outcome of using this tool enables engineers to prioritise maintenance for risk mitigation and

also provides decision-makers with a better understanding of the process of contaminant

intrusion.

It is anticipated that IRA-WDS will find wide application among water utilities, especially in

developing countries where intermittent water supplies are the norm.

References

gis projects for civil engineeing students

Documents